MR-Coupler: Automated Metamorphic Test Generation via Functional Coupling Analysis

2026-04-21 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Expert, extended

Summary

MR-Coupler is a novel automated approach designed to generate metamorphic test cases (MTCs) by analyzing functional coupling between methods within source code. This technique addresses the "oracle problem" in software testing, where expected outputs are difficult to specify. MR-Coupler operates in three phases: identifying functionally coupled method pairs based on signature commonality, function calls, and state interactions; using large language models (LLMs) to generate candidate MTCs with contextual guidance and refinement; and finally, amplifying and validating these MTCs through mutation analysis. Evaluation on 100 human-written MTCs and 50 real-world bugs demonstrated that MR-Coupler generates valid MTCs for over 90% of tasks, improving valid MTC generation by 64.90% and reducing false alarms by 36.56% compared to baselines. The generated MTCs also detected 44% of the real bugs, highlighting the approach's effectiveness and potential for practical adoption.

Key takeaway

For software engineers and QA teams struggling with the test oracle problem, MR-Coupler offers a practical solution to automate metamorphic test case generation. By leveraging functional coupling and LLMs, you can significantly reduce manual effort in creating effective metamorphic relations. Consider integrating this approach to enhance test adequacy and bug detection, especially for complex systems where expected outputs are hard to define, thereby improving your software's reliability.

Key insights

Functional coupling between methods can automate metamorphic relation construction and MTC generation, addressing the test oracle problem.

Principles

• Functional coupling is inferable from code.
• LLMs can generate MTCs with proper context.
• Mutation analysis validates MTCs without ground truth.

Method

MR-Coupler identifies coupled methods, prompts LLMs for MTC generation with invocation examples and templates, then refines and validates MTCs using input amplification and mutation analysis.

In practice

• Use signature commonality to find coupled methods.
• Provide LLMs with API usage examples for MTC generation.
• Validate LLM-generated MTCs via mutation analysis.

Topics

• Metamorphic Testing
• Metamorphic Relations
• MR-Coupler
• Functional Coupling Analysis
• Large Language Models

Code references

• diennea/herddb
• FasterXML/jackson-core
• apache/iotdb
• jcabi/jcabi-github
• locationtech/proj4j

Best for: AI Scientist, Research Scientist, Software Engineer

Related on AIssential

• STMutants: A Mutation Testing Dataset for Structured Text Programs in Industrial Automation — STMutants provides the first public mutation testing benchmark for IEC 61131-3 Structured Text, enabling reproducible research and LLM evaluation for PLC software.
• HYPERHEURIST: A Simulated Annealing-Based Control Framework for LLM-Driven Code Generation in Optimized Hardware Design — Decoupling functional correctness from PPA optimization stabilizes LLM-driven hardware design.
• Exploring LLM-based Verilog Code Generation with Data-Efficient Fine-Tuning and Testbench Automation — Automated testbench generation enables data-efficient LLM fine-tuning for high-quality Verilog code generation.
• MORTAR: Multi-turn Metamorphic Testing for LLM-based Dialogue Systems — MORTAR offers a bias-free, cost-effective metamorphic testing approach for multi-turn LLM dialogue systems.
• Meta Applies Mutation Testing with LLM to Improve Compliance Coverage — LLMs can significantly improve mutation testing by generating context-aware mutants and targeted tests, enhancing compliance coverage.
• Doubly Saturated Ramsey Graphs: A Case Study in Computer-Assisted Mathematical Discovery — Integrating SAT solving, LLMs, and formal verification accelerates mathematical discovery of complex graph structures.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.